Does early Computerised Tomography exclude fracture in ‘Clinical Scaphoid Fracture’?

1
Does early Computerised Tomography exclude
fracture in Clinical Scaphoid Fracture?

• Dr. Mark Harris
• Dr Jaycen Cruickshank
• Department of Orthopaedics, Emergency Medicine
  Radiology
• Ballarat Health Services

2
Why this research question?

• Unique blood supply of scaphoid vulnerable to
  non-union and AVN ? by delayed diagnosis
• Do we over-treat people to avoid missing it our
  fracture clinic felt many unnecessary patients
• On one occasion a difficult case solved with a
  CT, which prompted a literature review and
  research question.
• Insert trapezium fracture.

3
Scaphoid fracture

• Clinical examination
• Anatomical snuff box
• AP compression scaphoid
• Axial compression thumb
• All more sensitive than specific
• 10 to 25 not seen on initial radiographs
• Diagnostic uncertainty and fear of complications
  /litigation 10-14 days in plaster, with 10 of
  those patients subsequently having a scaphoid
• MRI has increased yield to 33, other fractures

4
Investigations

• Scaphoid XR series
• Cheap
• Available
• 1st line screening
• Low sensitivity on Day 1, and also at Day 10.

5
Investigations Bone Scan

• Murphy et al
• Day 4 Scan n100
• Sensitivity 100
• specificity 98????
• NPV 100 (97-100)
• PPV 66
• radiation
• Variable availability
• MBS 320.40
• Earlier less reliable

6
Investigations - MRI

• Compared to Bone Scan
• Sensitivity 100
• N 187, 37 scaphoid, 28 radius, 9 carpals 1
• Now gold standard
• More specific 2
• MBS 475
• Specialists only

1 Brydie Raby 2 Fowler et al, and Gabler et al
7
CT

• Compared to bone scan N29 1
• 100 NPV PPV
• Also some small studies, early late.
• MBS 190
• Accessible to GP/ED
• Newer multi-slice CTs

1 Breederveld and Tuinebreijer
8
Study design

• A prospective observational study of Emergency
  Department ED patients with clinical scaphoid
  fracture.
• Hypothesis
• Does early CT rule out fracture in clinical
  scaphoid fracture?
• Does early CT confirm fracture scaphoid or
  alternative diagnosis?
• Aim to avoid unnecessary plaster immobilisation
  in up to 9/10 patients

9
Study population

• Inclusion criteria
• A clinical scaphoid fracture will be defined as
  the presence of anatomical snuffbox tenderness
  in a patient with a mechanism of injury
  consistent with scaphoid trauma but normal
  initial radiographs.
• Exclusion criteria
• Patients under 18 years of age
• Patients who are known to be pregnant
• Patients who are unable to give informed consent

10
Study population (2)

• Ballarat Emergency Department
• Regional trauma centre
• 36,000 patients/yr, 27 paeds
• Recruitment by ED staff of varying seniority and
  experience.

11
Sample size Statistics.

• If the Ct is 90 sensitive and 95 specific, and
  the prevalence of scaphoid fracture is 10, the
  negative predictive value will be 99 in this
  patient group.
• We used a formula that calculates the standard
  error of a , relating it to the figure and
  sample size.

12
Assessment.

• The diagnostic value of an early computerized
  tomography CT scan will be validated compared
  to the gold standard diagnostic clinical
  protocol. Diagnosis of true scaphoid fracture is
  confirmed radiographically at the time of
  reassessment or, in the case of persistent
  tenderness and negative repeat radiographs, by
  Magnetic Resonance Imaging. This gold standard is
  similar to that used in studies using bone scans
  (Murphy and Eisenhauer 709-12)

13
Diagnostic Algorithm
14
CT results
15
CT vs Gold Standard Diagnosis
NPV 96.4 (92.1- 100) Sensitivity 87.5 PPV
100 Specificity 100
16
Discussion

• Prevalence of injuries() 43
• 7/35 occult scaphoid fractures (20)
• 8/35 other injuries (sensitivity specificity
  100)
• PPV 100 for scaphoid fracture and other
  fractures.
• NPV 96.4 for scaphoid fracture
• 1 False negative to discuss
• 1 patient had MRI reported NAD when CT
  unblinded, triquetral fracture review MRI then
  confirmed diagnosis.

17
Conclusion

• CT has a very high NPV for ruling out scaphoid
  fracture
• Continuing to a sample size of 100 will narrow
  the confidence interval
• CT good for excellent other injuries

18
External validity can we extrapolate?

• Groups similar to study population
• Similar patient groups and treating doctors are
  out there.
• In other populations
• Similar CT
• Prevalence 10, NPV to 98
• Larger sample may demostrate better sensitivity,
  and thus NPV also.
• In children?